The Conformational Properties of the Glc3Man Unit Suggest Conformational Biasing within the Chaperone-assisted Glycoprotein Folding Pathway

J. Mol. Biol. (2009) 387, 335–347

doi:10.1016/j.jmb.2009.01.043

Available online at www.sciencedirect.com

The Conformational Properties of the Glc3Man Unit Suggest Conformational Biasing within the Chaperone-assisted Glycoprotein Folding Pathway Mukram M. Mackeen 1 , Andrew Almond 2 , Michael Deschamps 3 , Ian Cumpstey 4 , Antony J. Fairbanks 4 , Clarence Tsang 1 , Pauline M. Rudd 1 , Terry D. Butters 1 , Raymond A. Dwek 1 and Mark R. Wormald 1 ⁎ 1

Oxford Glycobiology Institute, Department of Biochemistry, University of Oxford, South Parks Road, Oxford OX1 3QU, UK 2

Faculty of Life Sciences, University of Manchester, Manchester Interdisciplinary Biocenter, 131 Princess Street, Manchester M1 7DN, UK 3 Department of Biochemistry, University of Oxford, South Parks Road, Oxford OX1 3QU, UK 4

Chemistry Research Laboratory, Department of Chemistry, University of Oxford, South Parks Road, Oxford OX1 3TA, UK

A major puzzle is: are all glycoproteins routed through the ER calnexin pathway irrespective of whether this is required for their correct folding? Calnexin recognizes the terminal Glcα1–3Manα linkage, formed by trimming of the Glcα1–2Glcα1–3Glcα1–3Manα (Glc3Man) unit in Glc3Man9GlcNAc2. Different conformations of this unit have been reported. We have addressed this problem by studying the conformation of a series of Nglycans; i.e. Glc3ManOMe, Glc3Man4,5,7GlcNAc2 and Glc1Man9GlcNAc2 using 2D NMR NOESY, ROESY, T-ROESY and residual dipolar coupling experiments in a range of solvents, along with solution molecular dynamics simulations of Glc3ManOMe. Our results show a single conformation for the Glcα1–2Glcα and Glcα1–3Glcα linkages, and a major (65%) and a minor (30%) conformer for the Glcα1–3Manα linkage. Modeling of the binding of Glc1Man9GlcNAc2 to calnexin suggests that it is the minor conformer that is recognized by calnexin. This may be one of the mechanisms for controlling the rate of recruitment of proteins into the calnexin/calreticulin chaperone system and enabling proteins that do not require such assistance for folding to bypass the system. This is the first time evidence has been presented on glycoprotein folding that suggests the process may be optimized to balance the chaperone-assisted and chaperone-independent pathways. © 2009 Elsevier Ltd. All rights reserved.

Received 8 October 2008; received in revised form 19 January 2009; accepted 23 January 2009 Available online 29 January 2009 Edited by A. G. Palmer III

Keywords: glucosylated N-glycan conformation; glycoprotein folding; calnexin/calreticulin; glucosidase II; NMR

*Corresponding author. E-mail address: [email protected]. Current addresses: M. Deschamps, CRMHT-CNRS, 1D Avenue de la Recherche Scientifique, 45071 Orléans Cedex 2, France; I. Cumpstey, The Arrhenius Laboratory, Department of Organic Chemistry, Stockholm University, 106 91 Stockholm, Sweden; A. J. Fairbanks, Department of Chemistry, University of Canterbury, Christchurch 8140, New Zealand; P.M. Rudd, NIBRT, Conway Institute, University College Dublin, Belfield, Dublin 4, Ireland. Abbreviations used: ER, endoplasmic reticulum; OST, oligosaccharyl transfer complex; CDG, congenital disorders of glycosylation; MD, molecular dynamics; NOESY, nuclear Overhauser effect spectroscopy; ROE, rotating frame nuclear Overhauser effect; RDC, residual dipolar coupling; ROESY, rotating frame Overhauser effect spectroscopy; GDO, generalized degree of order. 0022-2836/$ - see front matter © 2009 Elsevier Ltd. All rights reserved.

336

Introduction The asparagine-linked glycosylation (N-glycosylation) of proteins is ubiquitous and takes place in the endoplasmic reticulum (ER) in association with the secretory pathway.1–4 Glucosylated oligomannose oligosaccharides have a crucial role during this process and in the subsequent protein folding and quality control processes in the ER.5,6 The oligosaccharyl transfer complex (OST) transfers the preformed oligosaccharide Glc3Man9GlcNAc2 (Fig. 1) co-translationally to the nascent polypeptide.7,8 OST preferentially utilizes the full-length triglucosylated glycan (Glc3Man9GlcNAc2) donor substrate over truncated forms.9,10 In humans, defects in the synthesis of the terminal glucose residues result in congenital disorders of glycosylation (CDG); for example, the deficient activity of α1–3 glucosyltransferases seen in CDG types 1c and 1 h cases.11 After co-translational transfer, Glc3Man9GlcNAc2 undergoes sequential trimming of the glucose residues by ER glucosidases I and II, to give Glc1Man9GlcNAc2. This glycan can either be processed further by glucosidase II or recognized by the ER-resident lectin chaperones calnexin (membranebound) and calreticulin (soluble) that facilitate protein folding and oligomerization by co-chaperones such as the oxidoreductase ERp57.12,13 The initial balance between these routes will depend on the relative rates of glucosidase II processing and chaperone binding. Chaperone-bound proteins are subsequently released after single or multiple binding cycles and the remaining glucose residue is removed by glucosidase II.14 Misfolded proteins are recognized by the enzyme UDP-glucose:glycoprotein glucosyl transferase, which reglucosylates their oligosaccharides, allowing the proteins to reenter the chaperone system for reprocessing.15,16 Proteins that are still misfolded after many cycles of glucose capping and uncapping are further processed by the ER resident mannosidase, 17 which removes the D2 mannose residue. This signals the protein to enter the ER-associated degradation (ERAD) pathway involving retrotranslocation from the ER to the cytosol and eventually to the proteasome.18

The Conformation of the Glc3Man Unit

Each of these steps is controlled by the oligosaccharide processing and involves specific recognition of the Glc3–0Man unit. Knowledge of the conformation of this unit is critical for understanding these recognition events. The synthetic fragment Glc3OMe was studied by molecular dynamics (MD) simulations in water and in vacuo and nuclear magnetic resonance (NMR) spectroscopy measurements of trans-glycosidic 3JCH coupling constants,19 suggesting that each linkage is flexible and the Glcα1–3Glcα linkage can adopt two possible conformations. The synthetic fragment Glc3ManOPr was studied by NMR spectroscopy in DMSO to identify stable hydrogen bonds and energy calculations in vacuo,20 which indicated a single rigid conformation for the fragment with a strong interaction between the first and third glucose residues. The conformation of the Glc3Man unit on Glc3Man7GlcNAc2 in water was derived from NMR nuclear Overhauser effect spectroscopy (NOESY) data,21 which again suggested a single rigid structure for the Glc3 unit but gave a different conformation for the Glcα1–2Glcα linkage. It is not clear if the differences between the results of these studies are due to differences in the sizes of the fragments used, differences in the solvents used, or inadequacies of either the experimental information available or the theoretical methods of calculation. To address these problems, we report here on the conformational studies of (i) Glc3ManOMe using NMR spectroscopy, both nuclear Overhauser effect (NOE)/rotating frame Overhauser effect (ROE) and residual dipolar coupling (RDC) experiments, in different solvents and using MD simulations in water; (ii) Glc3Man7GlcNAc2 using NMR spectroscopy, both NOE/ROE and RDC experiments, in different solvents; and (iii) Glc3Man4GlcNAc2 using NMR spectroscopy NOE/ROE experiments in different solvents and Glc3Man5GlcNAc2 and Glc1Man9GlcNAc2 using NMR spectroscopy NOE/ROE experiments in water. These results have been used to model the calnexin:Glc1Man9GlcNAc2 complex using the X-ray structure of calnexin:glucose and functional results of site-directed mutagenesis. Our results allowed us to resolve the discrepancy between the different conformations of the Glc3Man

Fig. 1. A representation of the Glc3Man9GlcNAc2 and Glc3ManOMe structures showing the primary sequence and residue numbering. Glc3Man7GlcNAc2 is missing residues D2 and D3; Glc3Man5GlcNAc2 is missing residues A, B, D2 and D3; Glc3Man4GlcNAc2 is missing residues 4′, A, B, D2 and D3; Glc1Man9GlcNAc2 is missing residues G1 and G2.

The Conformation of the Glc3Man Unit

unit reported earlier. We show that the Glc3 core in the Glc3Man unit is rigid but the subsequent GlcMan linkage in the unit is flexible with two conformations. The existence of two conformations for the Glc-Man linkage is of particular significance, since this linkage is recognized by both the molecular chaperones (calnexin/calreticulin) and glucosidase II. Analysis of the calnexin:glycan complex model suggests that binding to the minor conformer of the Glc-Man linkage is more consistent with the mutagenesis data reported earlier, with implications for the kinetics and thermodynamics of recruitment into the ER chaperone system.

Results Preparation, solubility and NMR assignment of oligosaccharides The structures of Glc3Man7GlcNAc2 and Glc3ManOMe are shown schematically in Fig. 1, together with the residue labeling scheme. Glc3ManOMe was synthesized, Glc3Man 7GlcNAc 2 and Glc1Man9GlcNAc2 were isolated from biological sources, and Glc3 Man 5 GlcNAc 2 and Glc 3 Man 4GlcNAc 2 were prepared by enzymatic digestion of Glc3Man7GlcNAc2, all as described in Materials and Methods. All oligosaccharides were N 90% pure, as shown by HPLC, mass spectrometry and NMR spectroscopy. All the oligosaccharides were soluble in both 2H2O and C2H3O2H, but only Glc3ManOMe was soluble in DMSO-d6 at concentrations sufficient to carry out NMR experiments. NMR resonance assignments were obtained by standard procedures for Glc3ManOMe in 2H2O, C2H3O2H and DMSO-d6, for Glc3Man7GlcNAc2 and Glc3Man4GlcNAc2 in 2H2O and C 2 H3O 2 H, and for Glc1Man9GlcNAc2 and Glc3Man5GlcNAc2 in 2H2O (given in the Supplementary Data). The assignments are fully consistent with values reported earlier.20–24

337 Solvent dependence of the Glc3Man unit conformation NOESY, rotating frame Overhauser effect spectroscopy (ROESY) and T-ROESY spectra were recorded on Glc3ManOMe in 2H2O, C2H3O2H and DMSO-d6, and on Glc3Man7GlcNAc2 and Glc3Man4GlcNAc2 in 2 H2O and C2H3O2H. The observed pattern of ROE peaks for the G1-G2, G2-G3 and G3-D1 linkages of any given oligosaccharide differed between different solvents (data not shown) but this was due to differences in local inter-proton correlation times in the different solvents. Once these effects had been allowed for, the same quantitative pattern of results was observed for each oligosaccharide in all solvents and the same inter-proton distance constraints obtained (as described below). Thus, the conformation of Glc3ManOMe appears to be the same in 2 H2O, C2H3O2H and DMSO-d6, and the conformation of the Glc3Man unit in Glc3Man7GlcNAc2 and Glc3Man4GlcNAc2 appears to be the same in 2H2O and C2H3O2H; i.e. the conformation of the Glc3Man unit is independent of the solvent. NMR determination of linkage conformations of Glc3ManOMe in water Cross-linkage inter-proton distance constraints, calculated allowing for variations in local correlation times,25 for the three glycosidic linkages in Glc3ManOMe were obtained from NOESY, ROESY and T-ROESY spectra. The anomeric region of the ROESY spectrum is shown in Fig. 2a and the interproton T-ROE intensities, local correlation times and calculated distances are given in Table 1. Two ROEs (and NOEs and T-ROEs) are seen across the G1-G2 (Glcα1–2Glcα) linkage. There is a single conformer (Φ = – 20° to + 20°, Ψ = – 40° to – 60°), which is consistent with the two distance constraints (torsion angle maps are given in the Supplementary Data). This conformer is consistent with the absence of all other ROEs. A conformer with Φ closer to + 20° is not consistent with the exo-anomeric effect.

Fig. 2. Anomeric region of the ROESY spectra in 2H2O, pH 8, 200 ms mixing time, of: (a) Glc3ManOMe; and (b) Glc3Man7GlcNAc2. Positive contours are in black, negative contours are in gray.

The Conformation of the Glc3Man Unit

338

Table 1. Measured correlation times (τij) and inter-proton distances from NMR spectroscopy and 〈r– 6〉– 1/6 averaged distances from the molecular dynamics (MD) simulation for proton pairs in Glc3ManOMe in 2H2O (for details, see Materials and Methods) Distance (Å) T-ROE intensity

τij (ps)

X-ray

NMRa

MD

G1-G2: Glcα1–2Glcα linkage G1:H1 G1:H2 G2:H1 G2:H2 G2:H3

7.36 8.28 5.30 –

280 289 200 –

2.34b

– 2.30 2.45 N3.5

2.41 ± 0.09 2.18 ± 0.12 2.76 ± 0.21 4.32 ± 0.13

G2-G3: Glcα1–3Glcα linkage G2:H1 G2:H2 G3:H2 G3:H3 G3:H4

7.12 – 10.65 –

338 – 316 –

2.34b 4.23 2.05 3.74

– N3.5 2.19 N3.5

2.42 ± 0.09 4.41 ± 0.12 2.55 ± 0.20 3.32 ± 0.23

G3-D1: Glcα1–3Manα linkage G3:H1 G3:H2 D1:H2 D1:H3 D1:H4

7.24c 1.22 12.72 ∼1.3d

347 172 260 nd

2.34b

– 3.08 2.12 ∼ 3.1

2.41 ± 0.09 3.48 ± 0.63 2.30 ± 0.18 3.69 ± 0.30

Proton pair

The X-ray distances are from the crystal structure of Glcα1–3GlcαOMe. nd – not determined. a An error of ±5% in measured NOE/ROEs intensities relative to the calibrant peak leads to an error in the calculated distances of ±0.02 Å for strong and ±0.03 Å for weak NOE/ROEs, respectively. An error of ±10% in measured NOE/ROE intensities results in errors of ±0.03 Å to ±0.06 Å in the calculated distances. b Used as the reference distance to calculate distances from T-ROE intensities. c Average of G1:H1-G1:H2 and G2:H1-G2:H2. d Partially overlapping, intensity obtained by subtraction of the G3:H1-G3:H2.

Only one ROE (and NOE and T-ROE) is seen across the G2-G3 (Glcα1–3Glcα) linkage, thus the linkage conformation is largely defined by the negative distance constraints from absent ROEs.26 There are two conformers consistent with both the observed and absent ROEs, one narrowly defined region of the torsion angle map at (Φ = – 25°, Ψ = – 20°) and the other much less well defined in the region (Φ = + 10° to + 50°, Ψ = –50° to +30°). Only the first of these is consistent with the exo-anomeric effect. The second region can be excluded because of unfavorable steric interactions between G2:O2 and G3:O4. The crystal structure of Glcα1–3GlcαOMe gives torsion angles of (Φ = – 18.2°, Ψ = – 24.9°).27 The agreement between the NMR distance constraints and the inter-proton distances from the crystal structure are excellent (Table 1). Three ROEs (and NOEs and T-ROEs) are seen across the G3-D1 (Glcα1–3Manα) linkage. Two of these, G3:H1-D1:H2 and G3:H1-D1:H4, are very weak (Table 1) and difficult to quantify, particularly the G3:H1-D1:H4 because of additional peak overlap. There is no single conformer consistent with all three of these distance constraints. The G3: H1-D1:H2 and G3:H1-D1:H3 constraints can be satisfied by a single sterically allowed conformer at (Φ = – 30°, Ψ = + 30°). To satisfy the G3:H1-D1:H4 constraint requires a conformer with Ψ b –40°. Thus, the NMR distance constraints are consistent with the G1-G2 and G2-G3 linkages adopting a single conformer, but the G3-D1 linkage must either adopt multiple distinct conformers or be very flexible. It is interesting to note that there is a greater variation in the local inter-proton correlation times

for the G3-D1 linkage than for the other linkages (Table 1), consistent with additional internal motion for this linkage. Molecular dynamics simulations of Glc3ManOMe in water The results of a 50 ns MD simulation of Glc3ManOMe in water are shown in Fig. 3 and the average 〈r- 6 〉 - 1/6 inter-proton distances for the observed ROEs are given in Table 1. The profiles of the conformational transitions suggested that conformational equilibration was reached for all the linkages over simulation period of 50 ns. The G1-G2 linkage adopts a single conformation for the entire length of the simulation, with torsion angles of (Φ = –55 ± 10°, Ψ = – 46 ± 11°) and approximately 100% occupancy. The G2-G3 linkage adopts a single defined conformation with torsion angles of (Φ = –53 ± 9°, Ψ = – 31 ± 11°) and 98% occupancy, with occasional changes to conformations with Ψ N 0. The G3-D1 linkage in contrast adopts two distinct conformers, a major conformer with torsion angles of (Φ = – 52 ± 11°, Ψ = –21 ± 11°) and 65% occupancy, and a minor conformer with torsion angles of (Φ = –46 ± 10°, Ψ = + 19 ± 12°) and 30% occupancy. The pattern of inter-nuclear distances observed in the MD simulation is very similar to those determined by NMR (and crystallography for the G2-G3 linkage), although the majority of the distances are slightly longer. The simulation also revealed that significant direct hydrogen bonding (N2% frequency) occurred only for G3:O4 to G1:O2 (10%), G1:O2 to G3:O4 (3%), D1:O4 to G3:O2 (5%), and

The Conformation of the Glc3Man Unit

339

Fig. 3. Results of a 50 ns molecular dynamics simulation of Glc3ManOMe in water. (a), (d) and (g), Plots of Φ torsion angle versus time; (b), (e) and (h), plots of Ψ torsion angle versus time, the boxes showing the distinct conformers. (c), (f) and (i), Φ/Ψ scatter plots; (a)–(c), G1-G2 linkage; (d)–(f), G2-G3 linkage; (g)–(i), G3-D1 linkage.

D1:O2 to G3:O5 (4%). Significant water-mediated hydrogen bonding was observed for D1:O4 to G3: O2 (33% monomer bridge and 21% dimer bridge).28 Residual dipolar coupling studies of Glc3ManOMe in water One-bond 1H-13C residual dipolar coupling constants (1DCH) were measured, at natural abundance using the 2D 1H-13C sensitivity-enhanced gradient experiment, proton-coupled in t1, from the difference in the peak splittings between the spectra of Glc3ManOMe recorded in the non-aligned (isotropic) and aligned (anisotropic) phases (Fig. 4). Filamentous bacteriophages29,30 were chosen as the alignment medium for several reasons: they are negatively charged at the surface and will therefore not interact on the basis of charge with the neutral N-glycans; direct recovery of the limited amounts of valuable N-glycans prepared for this study is possible by ultracentrifugation without recourse to chromatographic techniques, which would have resulted in significant loss of material; and the medium is stable in maintaining the alignment of the oligosaccharides within the desired temperature and pH ranges. The filamentous bacteriophages (Pfl) did not cause any significant change in 1H or 13C chemical shifts, or line-broadening in the heteronuclear single quantum coherence spectra, confirming that the oligosaccharides did not interact with the medium specifically. The solvent 2H spectrum showed sharp peaks of equal height to indicate a homogeneous preparation with quadrupolar splitting values of about 15 Hz.31

The experimental 1DCH values for Glc3ManOMe are given in the Supplementary Data. The range of observed 1DCH values is from – 10 Hz to + 11 Hz, and the values are predominantly positive. With oligosaccharides, the parallel orientations of C-Haxial vectors results in the less isotropic distribution of these vectors and is responsible for the observed “gaps” in the distribution of 1DCH values. The 1DCH values for the ring C-H vectors were used to validate the conformations obtained from the NOESY/ ROESY and MD data by comparing experimental and calculated 1DCH values, and to analyze internal motion using the generalized degree of order (GDO) parameter for each monosaccharide ring.32,33 The REDCAT software 34 was used to backcalculate the 1DCH values for both the major and minor conformations of Glc3ManOMe identified by the NOE analysis and MD simulations. Good agreement was found for both conformers, giving linear regression (R2) values of 0.96. The GDO values for the monosaccharide rings were calculated for both the major and minor conformers. These two conformers gave the same GDO values for each ring. The values for the major conformer are given in Table 2. The GDO parameter reflects the overall degree of ordering and will have identical values for rigidly connected molecular fragments. However, the presence of any internal motions on timescales faster than 1 ms between two fragments will attenuate the value of one fragment relative to the other.33 Residues G1 and G2 have identical GDO values, and residue G3 is of a similar magnitude. However, the GDO value for residue D1 is approximately an order of magnitude smaller.

The Conformation of the Glc3Man Unit

340

Fig. 4. t1-coupled, sensitivity-enhanced gradient, sensitivity-enhanced single-coherence quantum correlated spectra of the non-anomeric region of Glc3ManOMe in 2H2O, pH 8: (a) without Pf1 phages (isotropic); (b) with Pf1 phages (oriented); (c) expansion of the D1:C5-H5 peaks from the spectra in a and b showing the residual dipolar coupling. Positive contours are in black, negative contours are in gray.

Thus, the G1-G2 linkage is completely rigid, the G2G3 linkage is also reasonably rigid, and the G3-D1 linkage shows significant flexibility. As the RDC results show the G1-G2 and G2-G3 linkages to be rigid and thus adopt single conformations, the single conformations consistent with the NOE/ROE data, namely (Φ∼–20°, Ψ∼– 40°) for the G1-G2 linkage and (Φ = –25°, Ψ = – 20°) for the G2G3 linkage, must be the correct conformations. The NOE/ROE results required the G3-D1 linkage to be more flexible, as shown by the RDC results. The NOE/ROE and RDC NMR results are fully consistent with the results of the 50 ns MD simulation; Glc3ManOMe adopts two conformations, with relative populations of 65:30 (as shown in Fig. 5a and b, respectively), where the G1-G2-G3 fragment is reasonably rigid and the G3-D1 linkage is responsible for the conformational exchange. NMR studies of GlcxManyGlcNAc2 in water The pattern of ROEs observed for Glc3Man7GlcNAc2 (Table 3) was very similar to that of Glc3ManOMe. Glc3Man5GlcNAc2 and Glc3Man4GlcNAc2 show the same pattern of ROEs for the G1-G2, G2-G3 and G3-D1 linkages (data not shown). The pattern of NOEs is different for the G1-G2 linkage. At 500 MHz, only one NOE was observed for this linkage,21 the absence of any others being interpreted as a negative distance constraint. However, the second NOE can be seen at 700 MHz, and its absence at 500 MHz is due to local correlation time effects rather than a large inter-proton distance.25

Thus, negative distance constraints are used only when a peak is absent from the NOESY, ROESY and T-ROESY spectra. The pattern of NOEs for the G2-G3 linkage is very similar to the pattern of ROEs. In the original study of Glc3Man7GlcNAc2,21 only one NOE was observed for the G3-D1 linkage (mixing time 100 ms). An additional two NOEs are observed here, G3:H1-D1:H2 and G3:H1-D1:H4. They are both relatively weak, and one of them is partially obscured by a larger peak. Again, these three distance constraints are not consistent with a single conformation for this linkage. Glc1Man9GlcNAc2 is of particular interest because it is the ligand for calnexin and calreticulin. The Table 2. Generalized degree of order parameters calculated from the residual dipolar couplings for the Glc3Man residues in Glc3ManOMe and Glc3Man7GlcNAc2 (major conformers) Residue

GDOa

Glc3ManOMe G1 G2 G3 D1

1.37 × 10- 2 1.35 × 10- 2 3.04 × 10- 2 2.80 × 10- 3

Glc3Man7GlcNAc2 G1 G2 G3 D1

4.61 × 10- 3 4.40 × 10- 3 7.53 × 10- 2 1.27 × 10- 2

a RDC ±1 Hz was the estimated error used with the REDCAT software.33

The Conformation of the Glc3Man Unit

341

Fig. 5. Molecular models of the two conformations of Glc3ManOMe: (a) major conformer; (b) minor conformer. The positions of direct (dots close together) and water bridged (widely spaced dots) hydrogen bonds identified in the molecular dynamics simulation are shown.

NOESY spectrum at 500 MHz shows only one crosslinkage NOE for the G3-D1 linkage; namely, G3:H1D1:H3. However, the ROESY and T-ROESY spectra show a further two ROES, G3:H1-D1:H2 and G3:H1D1:H4, indicating that the lack of NOESY peaks is again due to differences in local correlation times. The pattern of ROEs is very similar to that of Glc3ManOMe, and the derived torsion angle map is virtually identical, indicating that G3-D1 linkage adopts multiple conformations in the absence of residues G1 and G2. One-bond 1H-13C residual dipolar coupling constants ( 1 D CH ) were measured for Glc 3 Man 7GlcNAc2, using filamentous bacteriophages as the alignment medium (values are given in the Supplementary Data). The range of observed 1DCH values is from – 15 Hz to + 23 Hz, indicating a greater degree of alignment for the larger oligosaccharide. The back-calculated 1DCH values using the two conformations of the Glc3Man unit as determined for Glc3ManOMe match the experimental data very well indeed (R2 = 0.99). The GDO values for the G1, G2, G3 and D1 residues are the same for each conformer and are given in Table 2. As for Glc3ManOMe, the G1-G2 linkage appears rigid and the G3-D1 linkage flexible. In contrast to the Glc3ManOMe, the G2-G3 linkage appears to be more flexible as well, although this does not affect the pattern of observed ROEs. Thus, the Glc3Man unit in Glc3Man7GlcNAc2 appears to have the same conformational behavior as determined by the pattern of ROEs as in Glc3ManOMe. The conformational behavior is unaffected by removal of the A, B and 4′ mannose residues, and the behavior of the G3-D1 linkage is also unaffected by removal of residues G1 and G2.

The flexibility of the G1-G2 and G3-D1 linkages is similar in Glc3ManOMe and Glc3Man7GlcNAc2, but the flexibility of the G2-G3 linkage increases in the latter. Table 3. Measured correlation times (τij) and distances from NMR spectroscopy for cross-linkage proton pairs in Glc 3 Man 7 GlcNAc 2 in 2 H 2O (see Materials in Supplementary Material for details) τij (ps)

Distance (Å)a

G1-G2: Glcα1–2Glcα linkage G1:H1 G1:H2 4.17 G2:H1 4.14 G2:H2 3.46 G2:H3 –

485 560 360 –

2.34b 2.35 2.41 N3.5

G2-G3: Glcα1–3Glcα linkage G2:H1 G2:H2 5.19 G3:H2 – G3:H3 5.82 G3:H4 –

743 – 821 –

2.34b N3.5 2.31 N3.5

G3-D1: Glcα1–3Manα linkage G3:H1 G3:H2 5.52 D1:H2 0.44 D1:H3 6.38 D1:H4 Weakc

776 678 926 nd

2.34b 3.55 2.31 ∼ 3.1

Proton pair

T-ROE intensity

nd – not determined. a An error of ±5% in measured NOE/ROEs intensities relative to the calibrant peak leads to an error in the calculated distances of ±0.02 Å for strong and ±0.03 Å for weak NOE/ROEs, respectively. An error of ±10% in measured NOE/ROE intensities results in errors of ±0.03 Å to ±0.06 Å in the calculated distances. b Reference value obtained from the crystal structure of Glcα1-3GlcαOMe. c The crosspeak could not be measured accurately due to overlap, but is approximately twice the size of the G3:H1-D1:H2 peak.

The Conformation of the Glc3Man Unit

342 Modeling of the complex between Glc1Man9GlcNAc2 and calnexin Molecular models of Glc1Man9GlcNAc2 were generated on the basis of the information above on the conformation of the G3-D1 linkage and studies reported earlier on the conformational studies of Man9GlcNAc2 by NMR and solvated MD.22 Two models were generated, one for the major conformer and one for the minor conformer of the G3-D1 linkage. Both oligosaccharides were docked to calnexin on the basis of the reported interactions between glucose and calnexin in the crystal structure.35 Crystallographic analysis of the calnexin: glucose complex showed significant interactions between the glucose and Met189, Tyr165, Lys167, Tyr186, Glu217 and Glu426 residues, which were later confirmed by site-directed mutagenesis.36 Docking of the Glc1Man9GlcNAc2 major conformer resulted in very severe steric interactions between residue D3 and B of the oligosaccharide and the protein (Fig. 6a). These could be relieved only by adjusting the 4′-3 Manα1–6Manβ linkage O6-C6-C5C4 torsion angle to 180°. However, these conformations lead to unfavorable steric interactions between residue D2, and residues 1 and 2, and they are not observed in Man9GlcNAc2 in solution.37 All of the other allowed Manα1–6Manα conformations38 gave steric clashes with the protein. Docking of the Glc1Man9GlcNAc2 minor conformer gave no unfavorable steric interactions with the protein (Fig. 6b). It resulted also in the close approach of residue C to Asp193, with a potential hydrogen bond between C: O4 and the carboxyl group. The equivalent residue to Asp193 in calreticulin (Asp135) has been identified by mutational studies as being critical to oligosaccharide binding. 39 Thus, binding of Glc1Man9GlcNAc2 with the minor conformer of the G3-D1 linkage appears to be sterically more plausible and fits better with the results of mutational studies.

Discussion The NMR NOE/ROE and RDC studies, and the MD simulations give a consistent conformational picture of Glc3ManOMe, showing that the G1-G2 and G2-G3 each adopt a single relatively rigid conformation and the G3-D1 linkage is more flexible and adopts two conformations. NMR studies show that these conformations are not solvent dependent. The rigidity of the G1-G2 and G2-G3 linkages is consistent with a direct hydrogen bond between residues G1 and G3 (Fig. 5). There is very little direct hydrogen bonding between G3 and D1. The most significant interaction is the water-bridged hydrogen bonding between D1:O4 and G3:O2, but this can be bridged by one or two water molecules. So it is not surprising that this linkage is more flexible. The only previous NMR study of Glc3ManOR used deuterium exchange experiments and the temperature dependence of the hydroxyl proton chemical shifts in DMSO to define the location of hydrogen bonds, combined with energy calculations using the HSEA forcefield.20 They identified a G1: O2 to G3:O4 direct hydrogen bond and found no evidence for a direct D1:O4 to G3:O2 hydrogen bond, consistent with the molecular dynamics simulation in water, and proposed a conformation with torsion angles of (Φ = –45°, Ψ = –30°) for the G1-G2 linkage, (Φ = –40°, Ψ = –20°) for the G2-G3 linkage, and (Φ = – 50°, Ψ = – 10°) for the G3-D1 linkage. This is very similar to the major conformer reported above. An MD study of Glc3OMe, supported by some NMR measurements, used five simulations with four different protocols.19 They observed a single major conformation of (Φ∼–55°, Ψ∼– 40°) for the G1-G2 linkage, although three of the simulations showed significant flexibility not observed here. They observed two different conformers for the G2-G3 linkage with torsion angles (Φ∼– 35°, Ψ∼+ 40°) in three simulations, two in

Fig. 6. Molecular models of the binding of Glc1Man9GlcNAc2 (blue) to calnexin (gray): (a) major conformer of G3-D1 linkage; (b) minor conformer of G3-D1 linkage. Residue Asp193 (equivalent to calreticulin Asp135, which has been shown by mutational studies to be critical to binding39) is shown in green. Residues in red have been shown to interact with the glucose moiety.35,36

The Conformation of the Glc3Man Unit

343

Fig. 7. Proposed scheme for chaperone-assisted protein folding. Glc3Man9GlcNAc2 is co-translationally attached to the unfolded peptide. The glycan is processed by glucosidases I and II (Glc I & II) to give Glc1Man9GlcNAc2. The major conformation of Glc1Man9GlcNAc2 is recognized by Glc II, which would allow the chaperone system to be bypassed, while calnexin recognizes the minor conformer and recruits the protein into the chaperone system. Whether a given protein is recruited initially into the chaperone system depends on the relative rates of these two steps. A relatively slow on rate to calnexin would increase the chance of a protein that folds quickly to avoid the chaperone system, and a relatively slow off rate from calnexin would ensure that proteins recruited into the chaperone system interact with the other chaperones. Correctly folded protein exits to the Golgi, while UDP-glucosyltransferase (UGT) reglucosylates incorrectly folded protein to return it to the cycle. The slower acting ER mannosidase (ER Man) removes the D2 mannose residue, producing Glc1/0Man8GlcNAc2, which targets the protein for the ER-associated degradation pathway (ERAD).

water and one in vacuo, and (Φ∼– 50°, Ψ∼–40°) in two simulations, both in vacuo. The latter two simulations in vacuo (using the forcefields CHEAT95 and HSEA) gave conformations that are very similar to those reported here, but again showed significant flexibility. NOE/ROE results in larger glucosylated oligomannose oligosaccharides indicate that the conformational properties of the Glc3Man unit are very similar to Glc3ManOMe. This is in contrast to the previous NOE study of Glc3Man7GlcNAc2,21 which suggested a significantly different conformation for the G1-G2 linkage based on the observation of only one crosslinkage NOE, as discussed above. The RDC results suggest that the G1-G2 linkage is still rigid but that the G2-G3 linkage becomes more flexible. If the direct hydrogen bond between G3:O4 and G1:O2 is significant in maintaining the rigidity of the G1-G2 and G2-G3 linkages in Glc3ManOMe, then this suggests that addition of extra mannose residues to the reducing terminus of Glc3Man affects this hydrogen bond. From the molecular model of Glc3Man7GlcNAc2, it is clear that there is no possible direct

interaction between any of the mannose residues and the groups involved in this hydrogen bond, so the only rationalization could be in terms of alterations of the hydration network surrounding these residues. Based on modeling, there is the potential for an interaction between G3:O6 and C:O3, which could mediate such a relayed effect. However, altering the Ψ torsion angle for the G2-G3 linkage has much less effect on the G3: O4 to G1:O2 distance than altering the Ψ torsion angle for the G1-G2 linkage, so the increased flexibility of the G2-G3 linkage may be accommodated whilst retaining the G1 to G3 hydrogen bond. Rigid glycosidic linkages have obvious advantages with respect to recognition by proteins, minimizing conformational enthalpy changes and reducing the loss of conformational entropy on binding. Thus, the observation of two conformations for the G3-D1 linkage has important functional consequences. This linkage is recognized both by the ER glucosidase II and by calnexin/calreticulin. It is most likely to adopt a single conformation when in complex with a protein, leading to a significant loss of configurational entropy even if it adopts the

344 minimum energy conformation in the complex. Glucosidase II cleaves both the G2-G3 and G3-D1 linkages but hydrolyses the latter at least an order of magnitude more slowly. As pointed out earlier,21 the major conformer of the G3-D1 disaccharide unit is almost identical with the conformation of the G2G3 disaccharide unit, the only difference being the epimerization at C2 of the non-reducing terminal residue. It is thus reasonable to assume that the major conformer of the G3-D1 linkage is required by glucosidase II. Even though this is the lowest energy conformation, it represents only 65% of population of this linkage. This may contribute to the reduced rate of hydrolysis compared to the G2-G3 linkage (population of 100%), although it could not explain the entire reduction. Modeling of the Glc1Man9GlcNAc2 complex with calnexin suggests that it is the minor conformer of the G3-D1 linkage that is required in this case for productive binding. This represents only 30% of the population of this linkage. As the interconversion between the two forms is fast, the binding of Glc1Man9GlcNAc2 to calnexin would shift the equilibrium towards the minor conformer potentially resulting in complete binding. However, binding to calnexin in vivo is in kinetic competition with glucosidase II processing and so selective recognition of two conformers may potentially alter the balance between the processes for folding glycoproteins (Fig. 7). The binding of the minor conformer of the substrate to calnexin would result in a reduction of the on-rate without necessarily affecting the off-rate (which depends on the breaking of the inter-molecular interactions in the bound state). A slower on-rate could allow more time for glucosidase II processing and folding to occur before the protein is recruited into the chaperone system, thus increasing the probability of rapidly folding proteins (or protein domains) bypassing the chaperone system entirely. Once bound to calnexin, a slow off-rate for the glycoprotein substrate would be beneficial, giving time for other chaperones (such as ERp57) to act on the misfolded glycoprotein in the complex. This control based on substrate conformational bias rather than direct substrate–protein interactions may thus be useful in increasing the efficiency of the protein folding process by reducing the amount of rapidly folding protein that is recruited into the chaperone machinery without decreasing the effectiveness of that machinery for proteins once they have been recruited. It would be interesting to see whether any such conformational biasing occurs during the ER-associated degradation pathway.

Materials and methods Preparation of oligosaccharides Glc3ManOMe was synthesized as described.40 Glc3Man7GlcNAc2 was purified by lectin affinity chromatography from CHO cells treated with the α-glucosidase inhibitor Nbutyl deoxynojirimicin as described.21

The Conformation of the Glc3Man Unit Glc3Man5GlcNAc2 and Glc3Man4GlcNAc2 were prepared by the enzymatic digestion of Glc3Man7GlcNAc2 with Jack bean α-mannosidase (for details, see the Supplementary Data). Glc1Man9GlcNAc2 was isolated from egg-yolk IgY antibody as described.41 but with slight modification using the EGGstract® IgY purification kit. The modified procedure involved identifying the positive fractions containing Glc1Man9GlcNAc2 in the unlabeled hydrazinolyzed material using MS and/or high-pH, anion-exchange chromatography, followed by separation on a preparative-scale TSK gel-filtration column. The purity of all oligosaccharides was checked by mass spectrometry, high-pH, anion-exchange chromatography and NMR spectroscopy. Analysis of oligosaccharides by mass spectrometry Matrix-assisted laser desorption/ionization mass spectrometry of oligosaccharides was carried out with a timeof-flight Finnigan LaserMat spectrometer using a 2,5dihydroxybenzoic acid matrix as described.42 Analysis of oligosaccharides by high-pH, anion-exchange chromatography The high-pH, anion-exchange chromatography with pulsed amperometric detection analysis of Glc3Man7GlcNAc2 and Glc3Man5GlcNAc2 fractions was carried out on a Dionex BioLC machine as described in the Supplementary Data. NMR spectroscopy The oligosaccharides samples were repeatedly dried, dissolved in either 500 μl of 2H2O, C2H3O2H or DMSO-d6 and transferred to either 5 mm NMR tubes or 5 mm Shigemi tubes with magnetic susceptibilities matched to 2 H2O or DMSO-d6. Chemical shifts were referenced to internal acetone (δH 2.225 ppm and δC 31.07 ppm) for 2 H2O, and to the residual signal of the methyl group in C2H3O2H and DMSO-d6 (δH 3.310 ppm and δH 2.500 ppm, respectively). Spectra were recorded on a Varian UnityINOVA 500 spectrometer, with a probe temperature of 30 °C. The 2D spectra were multiplied by sine or cosine-bell functions in both dimensions, as appropriate. 1H resonances were assigned from 2D phase-sensitive correlated spectroscopy, relayed correlated spectroscopy and/or total correlated spectroscopy spectra. Cross-linkage NOE/ROE patterns and comparison of reported assignments for glucosylated oligomannose-type oligosaccharides were used for sequence- and stereo-specific assignments.20–24 13 C resonances were assigned from 2D phase-sensitive heteronuclear multiple quantum coherence and/or heteronuclear single quantum coherence spectra. The 2D NOESY, ROESY and T-ROESY spectra were recorded for all samples. ROE build-up curves are shown in the Supplementary Data. Peak volume measurement, and the calculation of inter-proton correlation times and distance constraints were carried out as described before.25 The internal calibration used to calculate distances was the intra-residue H1-H2 NOE (or ROE). The distance for this proton-pair was measured for glucose from the crystal structure of Glcα1–3GlcαOMe,27 and for mannose from the crystal structure of Manα1–2Manα-OMe,43 both obtained from searching the

The Conformation of the Glc3Man Unit Cambridge Crystallographic Database44 at the Chemical Database Service at Daresbury.45 Residual dipolar coupling experiments The RDC experiments were carried out using filamentous phages (Pf1) to produce the anisotropic phase. Glc3ManOMe and Glc3Man7GlcNAc2 were concentrated to dryness, redissolved in 150 μl of 2H2O, and pipetted into a 1.5 ml Eppendorf tube. 10 mg of phages in 200 μl of 10 mM potassium phosphate buffer in 2H2O pH 8 was added to each sample and mixed by repeated pipetting. This was followed by centrifugation at 10,000 rpm in a table-top centrifuge to remove air bubbles, and the sugar-phage suspension wass transferred to a Shigemi tube with a magnetic susceptibility matched to 2H2O. Deuterium quadrupolar splitting was checked from 2H spectra acquired without proton decoupling. Residual dipolar coupling constants, 1DCH, were measured from the difference in splitting between the 2D 1H-13C t1coupled, gradient, sensitivity-enhanced single-coherence quantum correlated spectra of the samples in oriented (with phages) and isotropic (without phages) phases. The original 1H-decoupled gradient, sensitivity-enhanced single-coherence quantum correlated sequence available in the Varian library was modified to a t1-coupled sequence by the omission of the proton 180° pulse during the t1 evolution delay. Peak assignment and picking was carried out using the Sparky software†. The RDC data were processed using the REDCAT software34 based on order matrix analysis for well-defined structural fragments, namely the monosaccharide rings. The input is a coordinate file of a proposed structure and the experimental 1DCH values; the output is the principal axes of alignment, Euler angles, generalized degree of order GDO parameters for each monosaccharide ring and the back-calculated theoretical 1DCH values. Molecular dynamics simulation of Glc3ManOMe The conformation of Glc3ManOMe was simulated using solvated MD as described.28 The software CHARMM was used for the simulation using the GLYCAM carbohydrate force-field parameters46,47 modified for use with CHARMM. The TIP3P water model (three point charges) was used where the size of the central unit cell was 4.5 nm and the number of water molecules was 2141. Periodic boundary conditions tessellated in the face-centered cubic arrangement with the relevant Wigner–Seitz cell unit (a rhombic dodecahedron) was used to model water. The dielectric was maintained at the vacuum value (ɛ0) and non-bonded interactions were cut-off using the switching function between 0.8 nm and 1.2 nm. The simulation was run in the NVT ensemble (since no change in volume was expected) using the leap-frog modification of the Verlet algorithm with a time-step of 2 fs, and a temperature of 298 K by weak coupling to a heat-bath (coupling constant 5 ps− 1). Equilibration of 200 ps was run by strong coupling to a heath bath before the production dynamics run. Every five steps, non-bonded lists and images were updated frequently to maintain energy conservation. The simulation was performed at 25 °C for 50 ns and coordinates were output at 0.2 ps intervals. Hydrogen bonds were calculated as described.28 A hydrogen bond was defined to exist when the donor H to acceptor O distance was less than 0.35 nm † http://www.cgl.ucsf.edu/home/sparky/

345 and the angle OH–O was less than 60° from linear. The average occupancy of hydrogen-bonds was calculated for each interaction for every frame in each simulation. Only hydrogen bonds with a frequency higher than 2% were recorded. Hydrogen bonds involving water bridges were calculated in the same way and were counted when a water molecule was hydrogen bonded to more than two distinct polar groups. Water bridges with a frequency of less than 10% were discarded. Distinct conformers were identified as described.48 Torsion angles are quotes as mean ± standard deviation for each distinct conformer. Molecular modeling Molecular modeling was carried out on a Silicon Graphics Fuel workstation using the programs INSIGHT II and DISCOVER (Accelrys). Standard torsion angle nomenclature used throughout, i.e. for a 1–x linkage, ϕ = H1–C1–O–Cx′ and ψ = C1–O–Cx′–Hx′. Torsion angle maps were generated as described,49 making use of distance constraints from both the presence and the absence of ROEs.26 The bond lengths, angles and dihedral angles used to produce the maps were obtained from the disaccharide crystal structures of Glcα1–3Glcα-OMe, modifying the structure as necessary for the different linkages using INSIGHT II and DISCOVER. Molecular models of GlcxManyGlcNAc2 structures were generated by combining the structure of the Glc3Man unit and the results of previously reported conformational studies of Man9GlcNAc2.22 The crystal structure of calnexin35 was obtained from the PDB database.50

Acknowledgements The authors thank Luisa Fernandez-Guillen for assistance with HPLC, Brian Matthews for performing large-scale hydrazinolysis, Professor David Harvey for assistance with mass spectrometry, and Professor Frances Platt for providing fresh chicken eggs for the preparation of Glc1Man9GlcNAc2. This work was supported by funding from the Oxford Glycobiology Institute. The GLYCAM force-field effort is supported by Professor Robert Woods at the University of Georgia, and the force-field parameters can be downloaded from the website http:// glycam.ccrc.uga.edu/.

Supplementary Data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/ j.jmb.2009.01.043

References 1. Dwek, R. A. (1996). Glycobiology: toward understanding the function of sugars. Chem. Rev. 96, 683–720. 2. Roth, J. (2002). Protein N-glycosylation along the secretory pathway: relationship to organelle topography and function, protein quality control, and cell interactions. Chem. Rev. 102, 285–303.

346 3. Weerapana, E. & Imperiali, B. (2006). Asparaginelinked protein glycosylation: from eukaryotic to prokaryotic systems. Glycobiology, 16, 91R–101R. 4. Anelli, T. & Sitia, R. (2008). Protein quality control in the early secretory pathway. EMBO J. 27, 315–327. 5. Spiro, R. G. (2000). Glucose residues as key determinants in the biosynthesis and quality control of glycoproteins with N-linked oligosaccharides. J. Biol. Chem. 275, 35657–35660. 6. Trombetta, E. S. (2003). The contribution of N-glycans and their processing in the endoplasmic reticulum to glycoprotein biosynthesis. Glycobiology, 13, 77R–91R. 7. Kornfeld, S., Gregory, W. & Chapman, A. (1979). Class E Thy-1 negative mouse lymphoma cells utilize an alternate pathway of oligosaccharide processing to synthesize complex-type oligosaccharides. J. Biol. Chem. 254, 11649–11654. 8. Kornfeld, R. & Kornfeld, S. (1985). Assembly of asparagine-linked oligosaccharides. Annu. Rev. Biochem. 54, 631–664. 9. Trimble, R. B., Byrd, J. C. & Maley, F. (1980). Effect of glucosylation of lipid intermediates on oligosaccharide transfer in solubilized microsomes from Saccharomyces cerevisiae. J. Biol. Chem. 255, 11892–11895. 10. Karaoglu, D., Kelleher, D. J. & Gilmore, R. (2001). Allosteric regulation provides a molecular mechanism for preferential utilization of the fully assembled dolichol-linked oligosaccharide by the yeast oligosaccharyltransferase. Biochemistry, 40, 12193–12206. 11. Freeze, H. H. & Aebi, M. (2005). Altered glycan structures: the molecular basis of congenital disorders of glycosylation. Curr. Opin. Struct. Biol. 15, 448–490. 12. Williams, D. B. (2006). Beyond lectins: the calnexin/ calreticulin chaperone system of the endoplasmic reticulum. J. Cell Sci. 119, 615–623. 13. Caramelo, J. J. & Parodi, A. J. (2008). Getting in and out from calnexin/calreticulin cycles. J. Biol. Chem. 283, 10221–10225. 14. Solda, T., Galli, C., Kaufman, R. J. & Molinari, M. (2007). Substrate-specific requirements for UGT1-dependent release from calnexin. Mol. Cell, 27, 238–249. 15. Parodi, A. J. (2000). Protein glucosylation and its role in protein folding. Annu. Rev. Biochem. 69, 69–93. 16. Caramelo, J. J., Castro, O. A., de Prat Gay, G. & Parodi, A. J. (2004). The endoplasmic reticulum glucosyltransferase recognizes nearly native glycoprotein folding intermediates. J. Biol. Chem. 279, 46280–46285. 17. Herscovics, A. (2001). Structure and function of class I α1,2-mannosidases involved in glycoprotein synthesis and endoplasmic reticulum quality control. Biochimie, 83, 757–762. 18. Meusser, B., Hirsch, C., Jarosch, E. & Sommer, T. (2005). ERAD: the long road to destruction. Nature Cell Biol. 7, 766–772. 19. Hoog, C. & Widmalm, G. (2000). Molecular dynamics simulation and nuclear magnetic resonance studies of the terminal glucotriose unit found in the oligosaccharide of glycoprotein precursors. Arch. Biochem. Biophys. 377, 163–170. 20. Alvarado, E., Nukada, T., Ogawa, T. & Ballou, C. E. (1991). Conformation of the glucotriose unit in the lipid-linked oligosaccharide precursor for protein glycosylation. Biochemistry, 30, 881–886. 21. Petrescu, A. J., Butters, T. D., Reinkensmeier, G., Petrescu, S., Platt, F. M., Dwek, R. A. & Wormald, M. R. (1997). The solution NMR structure of glucosylated N-glycans involved in the early stages of glycoprotein biosynthesis and folding. EMBO J. 16, 4302–4310.

The Conformation of the Glc3Man Unit 22. Woods, R. J., Pathiaseril, A., Wormald, M. R., Edge, C. J. & Dwek, R. A. (1998). The high degree of internal flexibility observed for an oligomannose oligosaccharide does not alter the overall topology of the molecule. Eur. J. Biochem. 258, 372–386. 23. Di Patrizi, L., Capone, A., Focarelli, R., Rosati, F., Gallego, R. G., Gerwig, G. J. & Vliegenthart, J. F. (2001). Structural characterization of the N-glycans of gp273, the ligand for sperm-egg interaction in the mollusc bivalve Unio elongatulus. Glycoconjugate J. 18, 511–518. 24. Verostek, M. F., Atkinson, P. H. & Trimble, R. B. (1993). Glycoprotein biosynthesis in the alg3 Saccharomyces cerevisiae mutant. I. Role of glucose in the initial glycosylation of invertase in the endoplasmic reticulum. J. Biol. Chem. 268, 12095–12103. 25. Mackeen, M., Almond, A., Cumpstey, I., Enis, S. C., Kupce, E., Butters, T. D. et al. (2006). The importance of including local correlation times in the calculation of inter-proton distances from NMR measurements: ignoring local correlation times leads to significant errors in the conformational analysis of the Glcα1– 2Glcα linkage by NMR spectroscopy. Org. Biomol. Chem. 4, 2241–2246. 26. Wormald, M. R. & Edge, C. J. (1993). The systematic use of negative nuclear Overhauser constraints in the determination of oligosaccharide conformations: application to sialyl-Lewis X. Carbohydr. Res. 246, 337–344. 27. Neuman, A., Avenel, D., Arene, F., Gillier-Pandraud, H., Pougny, J. R. & Sinay, P. (1980). Crystal structure of methyl 3-O-α-D-glucopyranosyl-α-D-glucopyranoside (methyl α-nigeroside). Carbohydr. Res. 80, 15–24. 28. Almond, A. (2005). Towards understanding the interaction between oligosaccharides and water molecules. Carbohydr. Res. 340, 907–920. 29. Hansen, M. R., Mueller, L. & Pardi, A. (1998). Tunable alignment of macromolecules by filamentous phage yields dipolar coupling interactions. Nature Struct. Biol. 5, 1065–1074. 30. Prestegard, J. H. & Kishore, A. I. (2001). Partial alignment of biomolecules: an aid to NMR characterization. Curr. Opin. Chem. Biol. 5, 584–590. 31. Ottiger, M. & Bax, A. (1998). Characterization of magnetically oriented phospholipid micelles for measurement of dipolar couplings in macromolecules. J. Biomol. NMR, 12, 361–372. 32. Prestegard, J. H., al-Hashimi, H. M. & Tolman, J. R. (2000). NMR structures of biomolecules using field oriented media and residual dipolar couplings. Quart. Rev. Biophys. 33, 371–424. 33. Tian, F., Al-Hashimi, H. M., Craighead, J. L. & Prestegard, J. H. (2001). Conformational analysis of a flexible oligosaccharide using residual dipolar couplings. J. Am. Chem. Soc. 123, 485–492. 34. Valafar, H. & Prestegard, J. H. (2004). REDCAT: a residual dipolar coupling analysis tool. J. Magn. Reson. 167, 228–241. 35. Schrag, J. D., Bergeron, J. J., Li, Y., Borisova, S., Hahn, M., Thomas, D. Y. & Cygler, M. (2001). The structure of calnexin, an ER chaperone involved in quality control of protein folding. Mol. Cell, 8, 633–644. 36. Leach, M. R. & Williams, D. B. (2004). Lectindeficient calnexin is capable of binding class I histocompatibility molecules in vivo and preventing their degradation. J. Biol. Chem. 279, 9072–9079. 37. Wooten, E. W., Bazzo, R., Edge, C. J., Zamze, S., Dwek, R. A. & Rademacher, T. W. (1990). Primary sequence dependence of conformation in oligomannose oligosaccharides. Eur. Biophys. J. 18, 139–148.

The Conformation of the Glc3Man Unit 38. Wormald, M. R., Petrescu, A. J., Pao, Y. L., Glithero, A., Elliott, T. & Dwek, R. A. (2002). Conformational studies of oligosaccharides and glycopeptides: complementarity of NMR, X-ray crystallography, and molecular modelling. Chem. Rev. 102, 371–386. 39. Kapoor, M., Ellgaard, L., Gopalakrishnapai, J., Schirra, C., Gemma, E., Oscarson, S. et al. (2004). Mutational analysis provides molecular insight into the carbohydrate-binding region of calreticulin: pivotal roles of tyrosine-109 and aspartate-135 in carbohydrate recognition. Biochemistry, 43, 97–106. 40. Ennis, S. C., Cumpstey, I., Fairbanks, A. J., Butters, T. D., Mackeen, M. & Wormald, M. R. (2002). Total synthesis of the Glc3Man N-glycan tetrasaccharide. Tetrahedron, 58, 9403–9411. 41. Ohta, M., Hamako, J., Yamamoto, S., Hatta, H., Kim, M., Yamamoto, T. et al. (1991). Structures of asparagine-linked oligosaccharides from hen egg-yolk antibody (IgY). Occurrence of unusual glucosylated oligomannose type oligosaccharides in a mature glycoprotein. Glycoconj. J. 8, 400–413. 42. Harvey, D. J., Rudd, P. M., Bateman, R. H., Bordoli, R. S., Howes, K., Hoyes, J. B. & Vickers, R. G. (1994). Examination of complex oligosaccharides by matrixassisted laser desorption/ionization mass spectrometry on time-of-flight and magnetic sector instruments. Org. Mass Spectrom. 29, 753–765. 43. Srikrishnan, T., Chowdhary, M. S. & Matta, K. L. (1989). Crystal and molecular structure of methyl O-α-

347

44. 45. 46.

47.

48.

49.

50.

D-manno-pyranosyl-(1–2)-α-D-mannopyranoside. Carbohydr. Res. 186, 167–175. Allen, F. H. (2002). The Cambridge Structural Database: a quarter of a million crystal structures and rising. Acta Crystallogr. B, 58, 380–388. Fletcher, D. A., McMeeking, R. F. & Parkin, D. (1996). The United Kingdom Chemical Database Service. J. Chem. Inf. Comput. Sci. 36, 746–749. Woods, R. J., Dwek, R. A., Edge, C. J. & Fraser-Reid, B. (1995). Molecular mechanical and molecular dynamic simulations of glycoproteins and oligosaccharides. 1. GLYCAM_93 parameter development. J. Phys. Chem. 99, 3832–3846. Almond, A., Deangelis, P. L. & Blundell, C. D. (2006). Hyaluronan: the local solution conformation determined by NMR and computer modeling is close to a contracted left-handed 4-fold helix. J. Mol. Biol. 358, 1256–1269. Petrescu, A. J., Petrescu, S. M., Dwek, R. A. & Wormald, M. R. (1999). A statistical analysis of Nand O-glycan linkage conformations from crystallographic data. Glycobiology, 9, 343–352. Wooten, E. W., Edge, C. J., Bazzo, R., Dwek, R. A. & Rademacher, T. W. (1990). Uncertainties in structural determinations of oligosaccharide conformation, using measurements of nuclear Overhauser effects. Carbohydr. Res. 203, 13–17. Berman, H. M., Westbrook, J., Feng, Z., Gilliland, G., Bhat, T. N., Weissig, H. et al. (2000). The Protein Data Bank. Nucleic Acids Res. 28, 235–242.